From 5280a14a6079040205a1d968cd80f20448d047c7 Mon Sep 17 00:00:00 2001 From: Bart Van Assche Date: Tue, 15 Oct 2024 12:09:32 -0700 Subject: genirq: Introduce irq_get_nr_irqs() and irq_set_nr_irqs() Prepare for changing 'nr_irqs' from an exported global variable into a variable with file scope. This will prevent accidental changes of assignments to a local variable 'nr_irqs' into assignments to the global 'nr_irqs' variable. Suppose that a patch would be submitted for review that removes a declaration of a local variable with the name 'nr_irqs' and that that patch does not remove all assignments to that local variable. Such a patch converts an assignment to a local variable into an assignment into a global variable. If the 'nr_irqs' assignment is more than three lines away from other changes, the assignment won't be included in the diff context lines and hence won't be visible without inspecting the modified file. With these abstraction series applied, such accidental conversions from assignments to a local variable into an assignment to a global variable are converted into a compilation error. Signed-off-by: Bart Van Assche Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/all/20241015190953.1266194-2-bvanassche@acm.org --- include/linux/irqnr.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'include') diff --git a/include/linux/irqnr.h b/include/linux/irqnr.h index 3496baa0b07f..7419b807b71b 100644 --- a/include/linux/irqnr.h +++ b/include/linux/irqnr.h @@ -6,6 +6,8 @@ extern int nr_irqs; +unsigned int irq_get_nr_irqs(void) __pure; +unsigned int irq_set_nr_irqs(unsigned int nr); extern struct irq_desc *irq_to_desc(unsigned int irq); unsigned int irq_get_next_irq(unsigned int offset); -- cgit v1.2.3 From 1ad2048bf7146efb83bc033147ca1611a7fe8494 Mon Sep 17 00:00:00 2001 From: Bart Van Assche Date: Tue, 15 Oct 2024 12:09:52 -0700 Subject: genirq: Switch to irq_get_nr_irqs() Use the irq_get_nr_irqs() function instead of the global variable 'nr_irqs'. Cache the result of this function in a local variable in order not to rely on CSE (common subexpression elimination). Prepare for changing 'nr_irqs' from an exported global variable into a variable with file scope. Signed-off-by: Bart Van Assche Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/all/20241015190953.1266194-22-bvanassche@acm.org --- include/linux/irqnr.h | 33 +++++++++++++++++++-------------- kernel/irq/irqdomain.c | 2 +- kernel/irq/proc.c | 3 ++- 3 files changed, 22 insertions(+), 16 deletions(-) (limited to 'include') diff --git a/include/linux/irqnr.h b/include/linux/irqnr.h index 7419b807b71b..a33088d27c54 100644 --- a/include/linux/irqnr.h +++ b/include/linux/irqnr.h @@ -11,26 +11,31 @@ unsigned int irq_set_nr_irqs(unsigned int nr); extern struct irq_desc *irq_to_desc(unsigned int irq); unsigned int irq_get_next_irq(unsigned int offset); -# define for_each_irq_desc(irq, desc) \ - for (irq = 0, desc = irq_to_desc(irq); irq < nr_irqs; \ - irq++, desc = irq_to_desc(irq)) \ - if (!desc) \ - ; \ - else - +#define for_each_irq_desc(irq, desc) \ + for (unsigned int __nr_irqs__ = irq_get_nr_irqs(); __nr_irqs__; \ + __nr_irqs__ = 0) \ + for (irq = 0, desc = irq_to_desc(irq); irq < __nr_irqs__; \ + irq++, desc = irq_to_desc(irq)) \ + if (!desc) \ + ; \ + else # define for_each_irq_desc_reverse(irq, desc) \ - for (irq = nr_irqs - 1, desc = irq_to_desc(irq); irq >= 0; \ - irq--, desc = irq_to_desc(irq)) \ + for (irq = irq_get_nr_irqs() - 1, desc = irq_to_desc(irq); \ + irq >= 0; irq--, desc = irq_to_desc(irq)) \ if (!desc) \ ; \ else -# define for_each_active_irq(irq) \ - for (irq = irq_get_next_irq(0); irq < nr_irqs; \ - irq = irq_get_next_irq(irq + 1)) +#define for_each_active_irq(irq) \ + for (unsigned int __nr_irqs__ = irq_get_nr_irqs(); __nr_irqs__; \ + __nr_irqs__ = 0) \ + for (irq = irq_get_next_irq(0); irq < __nr_irqs__; \ + irq = irq_get_next_irq(irq + 1)) -#define for_each_irq_nr(irq) \ - for (irq = 0; irq < nr_irqs; irq++) +#define for_each_irq_nr(irq) \ + for (unsigned int __nr_irqs__ = irq_get_nr_irqs(); __nr_irqs__; \ + __nr_irqs__ = 0) \ + for (irq = 0; irq < __nr_irqs__; irq++) #endif diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c index e0bff21f30e0..ec6d8e72d980 100644 --- a/kernel/irq/irqdomain.c +++ b/kernel/irq/irqdomain.c @@ -1225,7 +1225,7 @@ int irq_domain_alloc_descs(int virq, unsigned int cnt, irq_hw_number_t hwirq, virq = __irq_alloc_descs(virq, virq, cnt, node, THIS_MODULE, affinity); } else { - hint = hwirq % nr_irqs; + hint = hwirq % irq_get_nr_irqs(); if (hint == 0) hint++; virq = __irq_alloc_descs(-1, hint, cnt, node, THIS_MODULE, diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index 9081ada81c3d..d226282c5b66 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -457,11 +457,12 @@ int __weak arch_show_interrupts(struct seq_file *p, int prec) } #ifndef ACTUAL_NR_IRQS -# define ACTUAL_NR_IRQS nr_irqs +# define ACTUAL_NR_IRQS irq_get_nr_irqs() #endif int show_interrupts(struct seq_file *p, void *v) { + const unsigned int nr_irqs = irq_get_nr_irqs(); static int prec; int i = *(loff_t *) v, j; -- cgit v1.2.3 From ef4c675dc2961ee533bdc1ea20390761df0af5be Mon Sep 17 00:00:00 2001 From: Bart Van Assche Date: Tue, 15 Oct 2024 12:09:53 -0700 Subject: genirq: Unexport nr_irqs Unexport nr_irqs and declare it static now that all code that reads or modifies nr_irqs has been converted to number_of_interrupts() / set_number_of_interrupts(). Change the type of 'nr_irqs' from 'int' into 'unsigned int' to match the return type and argument type of the irq_get_nr_iqs() / irq_set_nr_irqs() functions. Signed-off-by: Bart Van Assche Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/all/20241015190953.1266194-23-bvanassche@acm.org --- include/linux/irqnr.h | 1 - kernel/irq/irqdesc.c | 3 +-- 2 files changed, 1 insertion(+), 3 deletions(-) (limited to 'include') diff --git a/include/linux/irqnr.h b/include/linux/irqnr.h index a33088d27c54..e97206c721a0 100644 --- a/include/linux/irqnr.h +++ b/include/linux/irqnr.h @@ -5,7 +5,6 @@ #include -extern int nr_irqs; unsigned int irq_get_nr_irqs(void) __pure; unsigned int irq_set_nr_irqs(unsigned int nr); extern struct irq_desc *irq_to_desc(unsigned int irq); diff --git a/kernel/irq/irqdesc.c b/kernel/irq/irqdesc.c index b0733959f8ae..479cf1caa610 100644 --- a/kernel/irq/irqdesc.c +++ b/kernel/irq/irqdesc.c @@ -138,8 +138,7 @@ static void desc_set_defaults(unsigned int irq, struct irq_desc *desc, int node, desc_smp_init(desc, node, affinity); } -int nr_irqs = NR_IRQS; -EXPORT_SYMBOL_GPL(nr_irqs); +static unsigned int nr_irqs = NR_IRQS; /** * irq_get_nr_irqs() - Number of interrupts supported by the system. -- cgit v1.2.3 From 25caea955cc950507d179f3ef456404b475e8c23 Mon Sep 17 00:00:00 2001 From: Inochi Amaoto Date: Thu, 31 Oct 2024 14:08:58 +0800 Subject: irqchip: Add T-HEAD C900 ACLINT SSWI driver Add a driver for the T-HEAD C900 ACLINT SSWI device. This device allows the system with T-HEAD cpus to send ipi via fast device interface. Signed-off-by: Inochi Amaoto Signed-off-by: Thomas Gleixner Link: https://lore.kernel.org/all/20241031060859.722258-3-inochiama@gmail.com --- drivers/irqchip/Kconfig | 12 ++ drivers/irqchip/Makefile | 1 + drivers/irqchip/irq-thead-c900-aclint-sswi.c | 176 +++++++++++++++++++++++++++ include/linux/cpuhotplug.h | 1 + 4 files changed, 190 insertions(+) create mode 100644 drivers/irqchip/irq-thead-c900-aclint-sswi.c (limited to 'include') diff --git a/drivers/irqchip/Kconfig b/drivers/irqchip/Kconfig index ef0fa6975dab..9cac13663871 100644 --- a/drivers/irqchip/Kconfig +++ b/drivers/irqchip/Kconfig @@ -619,6 +619,18 @@ config STARFIVE_JH8100_INTC If you don't know what to do here, say Y. +config THEAD_C900_ACLINT_SSWI + bool "THEAD C9XX ACLINT S-mode IPI Interrupt Controller" + depends on RISCV + depends on SMP + select IRQ_DOMAIN_HIERARCHY + select GENERIC_IRQ_IPI_MUX + help + This enables support for T-HEAD specific ACLINT SSWI device + support. + + If you don't know what to do here, say Y. + config EXYNOS_IRQ_COMBINER bool "Samsung Exynos IRQ combiner support" if COMPILE_TEST depends on (ARCH_EXYNOS && ARM) || COMPILE_TEST diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile index 15ed6e952ecd..25e9ad29b8c4 100644 --- a/drivers/irqchip/Makefile +++ b/drivers/irqchip/Makefile @@ -103,6 +103,7 @@ obj-$(CONFIG_RISCV_APLIC_MSI) += irq-riscv-aplic-msi.o obj-$(CONFIG_RISCV_IMSIC) += irq-riscv-imsic-state.o irq-riscv-imsic-early.o irq-riscv-imsic-platform.o obj-$(CONFIG_SIFIVE_PLIC) += irq-sifive-plic.o obj-$(CONFIG_STARFIVE_JH8100_INTC) += irq-starfive-jh8100-intc.o +obj-$(CONFIG_THEAD_C900_ACLINT_SSWI) += irq-thead-c900-aclint-sswi.o obj-$(CONFIG_IMX_IRQSTEER) += irq-imx-irqsteer.o obj-$(CONFIG_IMX_INTMUX) += irq-imx-intmux.o obj-$(CONFIG_IMX_MU_MSI) += irq-imx-mu-msi.o diff --git a/drivers/irqchip/irq-thead-c900-aclint-sswi.c b/drivers/irqchip/irq-thead-c900-aclint-sswi.c new file mode 100644 index 000000000000..b0e366ade427 --- /dev/null +++ b/drivers/irqchip/irq-thead-c900-aclint-sswi.c @@ -0,0 +1,176 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright (C) 2024 Inochi Amaoto + */ + +#define pr_fmt(fmt) "thead-c900-aclint-sswi: " fmt +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define THEAD_ACLINT_xSWI_REGISTER_SIZE 4 + +#define THEAD_C9XX_CSR_SXSTATUS 0x5c0 +#define THEAD_C9XX_SXSTATUS_CLINTEE BIT(17) + +static int sswi_ipi_virq __ro_after_init; +static DEFINE_PER_CPU(void __iomem *, sswi_cpu_regs); + +static void thead_aclint_sswi_ipi_send(unsigned int cpu) +{ + writel_relaxed(0x1, per_cpu(sswi_cpu_regs, cpu)); +} + +static void thead_aclint_sswi_ipi_clear(void) +{ + writel_relaxed(0x0, this_cpu_read(sswi_cpu_regs)); +} + +static void thead_aclint_sswi_ipi_handle(struct irq_desc *desc) +{ + struct irq_chip *chip = irq_desc_get_chip(desc); + + chained_irq_enter(chip, desc); + + csr_clear(CSR_IP, IE_SIE); + thead_aclint_sswi_ipi_clear(); + + ipi_mux_process(); + + chained_irq_exit(chip, desc); +} + +static int thead_aclint_sswi_starting_cpu(unsigned int cpu) +{ + enable_percpu_irq(sswi_ipi_virq, irq_get_trigger_type(sswi_ipi_virq)); + + return 0; +} + +static int thead_aclint_sswi_dying_cpu(unsigned int cpu) +{ + thead_aclint_sswi_ipi_clear(); + + disable_percpu_irq(sswi_ipi_virq); + + return 0; +} + +static int __init thead_aclint_sswi_parse_irq(struct fwnode_handle *fwnode, + void __iomem *reg) +{ + struct of_phandle_args parent; + unsigned long hartid; + u32 contexts, i; + int rc, cpu; + + contexts = of_irq_count(to_of_node(fwnode)); + if (!(contexts)) { + pr_err("%pfwP: no ACLINT SSWI context available\n", fwnode); + return -EINVAL; + } + + for (i = 0; i < contexts; i++) { + rc = of_irq_parse_one(to_of_node(fwnode), i, &parent); + if (rc) + return rc; + + rc = riscv_of_parent_hartid(parent.np, &hartid); + if (rc) + return rc; + + if (parent.args[0] != RV_IRQ_SOFT) + return -ENOTSUPP; + + cpu = riscv_hartid_to_cpuid(hartid); + + per_cpu(sswi_cpu_regs, cpu) = reg + i * THEAD_ACLINT_xSWI_REGISTER_SIZE; + } + + pr_info("%pfwP: register %u CPU%s\n", fwnode, contexts, str_plural(contexts)); + + return 0; +} + +static int __init thead_aclint_sswi_probe(struct fwnode_handle *fwnode) +{ + struct irq_domain *domain; + void __iomem *reg; + int virq, rc; + + /* If it is T-HEAD CPU, check whether SSWI is enabled */ + if (riscv_cached_mvendorid(0) == THEAD_VENDOR_ID && + !(csr_read(THEAD_C9XX_CSR_SXSTATUS) & THEAD_C9XX_SXSTATUS_CLINTEE)) + return -ENOTSUPP; + + if (!is_of_node(fwnode)) + return -EINVAL; + + reg = of_iomap(to_of_node(fwnode), 0); + if (!reg) + return -ENOMEM; + + /* Parse SSWI setting */ + rc = thead_aclint_sswi_parse_irq(fwnode, reg); + if (rc < 0) + return rc; + + /* If mulitple SSWI devices are present, do not register irq again */ + if (sswi_ipi_virq) + return 0; + + /* Find riscv intc domain and create IPI irq mapping */ + domain = irq_find_matching_fwnode(riscv_get_intc_hwnode(), DOMAIN_BUS_ANY); + if (!domain) { + pr_err("%pfwP: Failed to find INTC domain\n", fwnode); + return -ENOENT; + } + + sswi_ipi_virq = irq_create_mapping(domain, RV_IRQ_SOFT); + if (!sswi_ipi_virq) { + pr_err("unable to create ACLINT SSWI IRQ mapping\n"); + return -ENOMEM; + } + + /* Register SSWI irq and handler */ + virq = ipi_mux_create(BITS_PER_BYTE, thead_aclint_sswi_ipi_send); + if (virq <= 0) { + pr_err("unable to create muxed IPIs\n"); + irq_dispose_mapping(sswi_ipi_virq); + return virq < 0 ? virq : -ENOMEM; + } + + irq_set_chained_handler(sswi_ipi_virq, thead_aclint_sswi_ipi_handle); + + cpuhp_setup_state(CPUHP_AP_IRQ_THEAD_ACLINT_SSWI_STARTING, + "irqchip/thead-aclint-sswi:starting", + thead_aclint_sswi_starting_cpu, + thead_aclint_sswi_dying_cpu); + + riscv_ipi_set_virq_range(virq, BITS_PER_BYTE); + + /* Announce that SSWI is providing IPIs */ + pr_info("providing IPIs using THEAD ACLINT SSWI\n"); + + return 0; +} + +static int __init thead_aclint_sswi_early_probe(struct device_node *node, + struct device_node *parent) +{ + return thead_aclint_sswi_probe(&node->fwnode); +} +IRQCHIP_DECLARE(thead_aclint_sswi, "thead,c900-aclint-sswi", thead_aclint_sswi_early_probe); diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index 2361ed4d2b15..799052249c7b 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -147,6 +147,7 @@ enum cpuhp_state { CPUHP_AP_IRQ_EIOINTC_STARTING, CPUHP_AP_IRQ_AVECINTC_STARTING, CPUHP_AP_IRQ_SIFIVE_PLIC_STARTING, + CPUHP_AP_IRQ_THEAD_ACLINT_SSWI_STARTING, CPUHP_AP_IRQ_RISCV_IMSIC_STARTING, CPUHP_AP_IRQ_RISCV_SBI_IPI_STARTING, CPUHP_AP_ARM_MVEBU_COHERENCY, -- cgit v1.2.3 From 49a17639508c3b35f90ca829e60dddeeeb750e74 Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Wed, 6 Nov 2024 15:51:39 +0100 Subject: softirq: Use a dedicated thread for timer wakeups on PREEMPT_RT. The timer and hrtimer soft interrupts are raised in hard interrupt context. With threaded interrupts force enabled or on PREEMPT_RT this leads to waking the ksoftirqd for the processing of the soft interrupt. ksoftirqd runs as SCHED_OTHER task which means it will compete with other tasks for CPU resources. This can introduce long delays for timer processing on heavy loaded systems and is not desired. Split the TIMER_SOFTIRQ and HRTIMER_SOFTIRQ processing into a dedicated timers thread and let it run at the lowest SCHED_FIFO priority. Wake-ups for RT tasks happen from hardirq context so only timer_list timers and hrtimers for "regular" tasks are processed here. The higher priority ensures that wakeups are performed before scheduling SCHED_OTHER tasks. Using a dedicated variable to store the pending softirq bits values ensure that the timer are not accidentally picked up by ksoftirqd and other threaded interrupts. It shouldn't be picked up by ksoftirqd since it runs at lower priority. However if ksoftirqd is already running while a timer fires, then ksoftird will be PI-boosted due to the BH-lock to ktimer's priority. The timer thread can pick up pending softirqs from ksoftirqd but only if the softirq load is high. It is not be desired that the picked up softirqs are processed at SCHED_FIFO priority under high softirq load but this can already happen by a PI-boost by a force-threaded interrupt. [ frederic@kernel.org: rcutorture.c fixes, storm fix by introduction of local_timers_pending() for tick_nohz_next_event() ] [ junxiao.chang@intel.com: Ensure ktimersd gets woken up even if a softirq is currently served. ] Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Thomas Gleixner Reviewed-by: Paul E. McKenney [rcutorture] Reviewed-by: Frederic Weisbecker Link: https://lore.kernel.org/all/20241106150419.2593080-4-bigeasy@linutronix.de --- include/linux/interrupt.h | 47 ++++++++++++++++++++++++++++++++ kernel/rcu/rcutorture.c | 8 ++++++ kernel/softirq.c | 69 ++++++++++++++++++++++++++++++++++++++++++++++- kernel/time/hrtimer.c | 4 +-- kernel/time/tick-sched.c | 2 +- kernel/time/timer.c | 2 +- 6 files changed, 127 insertions(+), 5 deletions(-) (limited to 'include') diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h index 457151f9f263..8cd9327e4e78 100644 --- a/include/linux/interrupt.h +++ b/include/linux/interrupt.h @@ -616,6 +616,53 @@ extern void __raise_softirq_irqoff(unsigned int nr); extern void raise_softirq_irqoff(unsigned int nr); extern void raise_softirq(unsigned int nr); +/* + * With forced-threaded interrupts enabled a raised softirq is deferred to + * ksoftirqd unless it can be handled within the threaded interrupt. This + * affects timer_list timers and hrtimers which are explicitly marked with + * HRTIMER_MODE_SOFT. + * With PREEMPT_RT enabled more hrtimers are moved to softirq for processing + * which includes all timers which are not explicitly marked HRTIMER_MODE_HARD. + * Userspace controlled timers (like the clock_nanosleep() interface) is divided + * into two categories: Tasks with elevated scheduling policy including + * SCHED_{FIFO|RR|DL} and the remaining scheduling policy. The tasks with the + * elevated scheduling policy are woken up directly from the HARDIRQ while all + * other wake ups are delayed to softirq and so to ksoftirqd. + * + * The ksoftirqd runs at SCHED_OTHER policy at which it should remain since it + * handles the softirq in an overloaded situation (not handled everything + * within its last run). + * If the timers are handled at SCHED_OTHER priority then they competes with all + * other SCHED_OTHER tasks for CPU resources are possibly delayed. + * Moving timers softirqs to a low priority SCHED_FIFO thread instead ensures + * that timer are performed before scheduling any SCHED_OTHER thread. + */ +DECLARE_PER_CPU(struct task_struct *, ktimerd); +DECLARE_PER_CPU(unsigned long, pending_timer_softirq); +void raise_ktimers_thread(unsigned int nr); + +static inline unsigned int local_timers_pending_force_th(void) +{ + return __this_cpu_read(pending_timer_softirq); +} + +static inline void raise_timer_softirq(unsigned int nr) +{ + lockdep_assert_in_irq(); + if (force_irqthreads()) + raise_ktimers_thread(nr); + else + __raise_softirq_irqoff(nr); +} + +static inline unsigned int local_timers_pending(void) +{ + if (force_irqthreads()) + return local_timers_pending_force_th(); + else + return local_softirq_pending(); +} + DECLARE_PER_CPU(struct task_struct *, ksoftirqd); static inline struct task_struct *this_cpu_ksoftirqd(void) diff --git a/kernel/rcu/rcutorture.c b/kernel/rcu/rcutorture.c index bb75dbf5c800..270c31a1e857 100644 --- a/kernel/rcu/rcutorture.c +++ b/kernel/rcu/rcutorture.c @@ -2440,6 +2440,14 @@ static int rcutorture_booster_init(unsigned int cpu) WARN_ON_ONCE(!t); sp.sched_priority = 2; sched_setscheduler_nocheck(t, SCHED_FIFO, &sp); +#ifdef CONFIG_IRQ_FORCED_THREADING + if (force_irqthreads()) { + t = per_cpu(ktimerd, cpu); + WARN_ON_ONCE(!t); + sp.sched_priority = 2; + sched_setscheduler_nocheck(t, SCHED_FIFO, &sp); + } +#endif } /* Don't allow time recalculation while creating a new task. */ diff --git a/kernel/softirq.c b/kernel/softirq.c index d082e7840f88..7b525c904462 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -624,6 +624,24 @@ static inline void tick_irq_exit(void) #endif } +#ifdef CONFIG_IRQ_FORCED_THREADING +DEFINE_PER_CPU(struct task_struct *, ktimerd); +DEFINE_PER_CPU(unsigned long, pending_timer_softirq); + +static void wake_timersd(void) +{ + struct task_struct *tsk = __this_cpu_read(ktimerd); + + if (tsk) + wake_up_process(tsk); +} + +#else + +static inline void wake_timersd(void) { } + +#endif + static inline void __irq_exit_rcu(void) { #ifndef __ARCH_IRQ_EXIT_IRQS_DISABLED @@ -636,6 +654,10 @@ static inline void __irq_exit_rcu(void) if (!in_interrupt() && local_softirq_pending()) invoke_softirq(); + if (IS_ENABLED(CONFIG_IRQ_FORCED_THREADING) && force_irqthreads() && + local_timers_pending_force_th() && !(in_nmi() | in_hardirq())) + wake_timersd(); + tick_irq_exit(); } @@ -971,12 +993,57 @@ static struct smp_hotplug_thread softirq_threads = { .thread_comm = "ksoftirqd/%u", }; +#ifdef CONFIG_IRQ_FORCED_THREADING +static void ktimerd_setup(unsigned int cpu) +{ + /* Above SCHED_NORMAL to handle timers before regular tasks. */ + sched_set_fifo_low(current); +} + +static int ktimerd_should_run(unsigned int cpu) +{ + return local_timers_pending_force_th(); +} + +void raise_ktimers_thread(unsigned int nr) +{ + trace_softirq_raise(nr); + __this_cpu_or(pending_timer_softirq, BIT(nr)); +} + +static void run_ktimerd(unsigned int cpu) +{ + unsigned int timer_si; + + ksoftirqd_run_begin(); + + timer_si = local_timers_pending_force_th(); + __this_cpu_write(pending_timer_softirq, 0); + or_softirq_pending(timer_si); + + __do_softirq(); + + ksoftirqd_run_end(); +} + +static struct smp_hotplug_thread timer_thread = { + .store = &ktimerd, + .setup = ktimerd_setup, + .thread_should_run = ktimerd_should_run, + .thread_fn = run_ktimerd, + .thread_comm = "ktimers/%u", +}; +#endif + static __init int spawn_ksoftirqd(void) { cpuhp_setup_state_nocalls(CPUHP_SOFTIRQ_DEAD, "softirq:dead", NULL, takeover_tasklets); BUG_ON(smpboot_register_percpu_thread(&softirq_threads)); - +#ifdef CONFIG_IRQ_FORCED_THREADING + if (force_irqthreads()) + BUG_ON(smpboot_register_percpu_thread(&timer_thread)); +#endif return 0; } early_initcall(spawn_ksoftirqd); diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c index 5402e0f24217..d9911516e743 100644 --- a/kernel/time/hrtimer.c +++ b/kernel/time/hrtimer.c @@ -1811,7 +1811,7 @@ retry: if (!ktime_before(now, cpu_base->softirq_expires_next)) { cpu_base->softirq_expires_next = KTIME_MAX; cpu_base->softirq_activated = 1; - __raise_softirq_irqoff(HRTIMER_SOFTIRQ); + raise_timer_softirq(HRTIMER_SOFTIRQ); } __hrtimer_run_queues(cpu_base, now, flags, HRTIMER_ACTIVE_HARD); @@ -1906,7 +1906,7 @@ void hrtimer_run_queues(void) if (!ktime_before(now, cpu_base->softirq_expires_next)) { cpu_base->softirq_expires_next = KTIME_MAX; cpu_base->softirq_activated = 1; - __raise_softirq_irqoff(HRTIMER_SOFTIRQ); + raise_timer_softirq(HRTIMER_SOFTIRQ); } __hrtimer_run_queues(cpu_base, now, flags, HRTIMER_ACTIVE_HARD); diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c index 753a184c7090..976a212cca2e 100644 --- a/kernel/time/tick-sched.c +++ b/kernel/time/tick-sched.c @@ -859,7 +859,7 @@ static void tick_nohz_restart(struct tick_sched *ts, ktime_t now) static inline bool local_timer_softirq_pending(void) { - return local_softirq_pending() & BIT(TIMER_SOFTIRQ); + return local_timers_pending() & BIT(TIMER_SOFTIRQ); } /* diff --git a/kernel/time/timer.c b/kernel/time/timer.c index 1759de934284..06f0bc1db6d9 100644 --- a/kernel/time/timer.c +++ b/kernel/time/timer.c @@ -2499,7 +2499,7 @@ static void run_local_timers(void) */ if (time_after_eq(jiffies, READ_ONCE(base->next_expiry)) || (i == BASE_DEF && tmigr_requires_handle_remote())) { - __raise_softirq_irqoff(TIMER_SOFTIRQ); + raise_timer_softirq(TIMER_SOFTIRQ); return; } } -- cgit v1.2.3